Terminal uridyltransferase 7 regulates TLR4-triggered inflammation by controlling Regnase-1 mRNA uridylation and degradation

Different levels of regulatory mechanisms, including posttranscriptional regulation, are needed to elaborately regulate inflammatory responses to prevent harmful effects. Terminal uridyltransferase 7 (TUT7) controls RNA stability by adding uridines to its 3′ ends, but its function in innate immune response remains obscure. Here we reveal that TLR4 activation induces TUT7, which in turn selectively regulates the production of a subset of cytokines, including Interleukin 6 (IL-6). TUT7 regulates IL-6 expression by controlling ribonuclease Regnase-1 mRNA (encoded by Zc3h12a gene) stability. Mechanistically, TLR4 activation causes TUT7 to bind directly to the stem-loop structure on Zc3h12a 3′-UTR, thereby promotes Zc3h12a uridylation and degradation. Zc3h12a from LPS-treated TUT7-sufficient macrophages possesses increased oligo-uridylated ends with shorter poly(A) tails, whereas oligo-uridylated Zc3h12a is significantly reduced in Tut7-/- cells after TLR4 activation. Together, our findings reveal the functional role of TUT7 in sculpting TLR4-driven responses by modulating mRNA stability of a selected set of inflammatory mediators.

I nflammatory cytokines and chemokines are mediators of inflammation in the innate arm of immunity. Innate cells produce inflammatory cytokines and chemokines rapidly upon infection or tissue injury through an array of patternrecognition receptors (PRRs) 1 . Inappropriate expression of inflammatory cytokines dysregulates inflammation, causing numerous disorders, including inflammatory and autoimmune diseases. Therefore, orchestrating the expression of inflammatory cytokines for an effective inflammatory response without harmful immunopathology is an important issue.
Recognition of conserved pathogen-associated molecular patterns (PAMPs) on microorganisms and damage-associated molecular patterns (DAMPs) released from stressed or injured cells by PRRs initiates pro-inflammatory cytokine responses and coordinates adaptive immune responses 2 . Amongst all PRR families, Toll-like receptors (TLRs) are the most wellcharacterized and they play an important role in innate immune responses 2 . TLR4 recognizes lipopolysaccharides (LPS) on Gram-negative bacteria 3 . Upon engagement with its ligand, TLR4 recruits Myeloid differentiated primary response gene-88 (MyD88) and Toll/IL-1 receptor domain-containing adapterinducing IFNβ (TRIF) to activate inhibitor of transcription factors NF-κB kinase (IKK)/NF-κB and TANK-binding kinase 1 (TBK1)/interferon regulatory factor 3 (IRF3) for the production of pro-inflammatory cytokines/chemokines and type I interferons, respectively 4,5 . Inflammatory responses induced by TLRs or other PRRs are controlled by layers of regulatory mechanisms including: modulation of signaling transduction, posttranscriptional regulation (PTR), and posttranslational modification (PTM) 6,7 . PTR which includes RNA splicing, editing, and decay, is a quick and effective way to control the quantity of mRNA 2,[8][9][10] . The role of PTR in regulation of PRR-triggered inflammatory response has come to light as of late.
Many PRR-triggered cytokine mRNAs have short half-lives which allows rapid control of cytokine production 6 . Cytokine mRNAs are subject to various regulatory cis-elements in their 3′-UTRs, such as microRNA (miRNA) target sites, AU-rich elements (AREs), and stem-loop structures. Many RNA-binding proteins (RBPs), including Regnase-1 11 , Arid5a 12 , Tristetraprolin (TTP) 13 , and AU-rich element RNA-binding protein 1 (AUF-1) 14 , specifically bind regulatory RNA cis-elements to control the stability of cytokine mRNAs. TTP and AUF1 recognize AREs in the 3′-UTR of numerous cytokines, including tumor necrosis factor (TNF) and IL-6, to initiate their decay 13,14 . The ribonuclease (RNase) Regnase-1 (also called Mcpip1 encoded by the Zc3h12a gene) modulates IL-6 mRNA stability when macrophages are activated by LPS or IL-1β 15 . Mechanistic studies revealed that Regnase-1 targets the stem-loop structure on Il6 3′-UTR and cleaves Il6 mRNA through cooperation with RNA helicase protein, UPF1 11,16 . On the other hand, stimulation with LPS and cytokines also enhances the expression of Arid5a. The presence of Arid5a impedes Regnase-1 binding to Il6 3′-UTR thereby stabilizes Il6 mRNA 12 . Therefore, it appears that regulation of spatiotemporal expression of RBPs can fine tune the quantity of cytokine mRNAs during inflammation.
Recent studies show that modification of RNAs on their 3′ ends by non-templated nucleotide addition is an evolutionary conserved mechanism for the control of RNA stability and its fate 17 . Among the known template-independent RNA 3′ terminal modifications, polyadenylation and uridylation are the two most studied mechanisms 18 . Uridylation is catalyzed by terminal uridyltransferases (TUTs). TUTs belong to a family of noncanonical poly(A) polymerases. Their function in RNA processing is evolutionally conserved from Schizosaccharomyces pombe to human 19 . TUT4 (also known as ZCCHC11) and TUT7 (also named ZCCHC6) are primarily responsible for cytoplasmic 3′ uridylation 20,21 of various RNAs, including precursor microRNAs (pre-miRNAs) 22,23 , mature miRNAs 24 , histone mRNAs 25,26 , cellular mRNAs 20 , noncoding RNAs 27 , and viral RNAs 28 in mammalian cells. Catalytic activities of TUT4/7 can monouridylate or oligo-uridylate pre-miRNA resulting in biogenesis or degradation of miRNAs, respectively 21,23 . TUT4/7 are also known to trigger oligo-uridylation of poly(A)-tail-lacking histone mRNA at the end of S phase of a cell cycle to enhance its degradation 25,26,29 . Recently, TAIL-seq analysis revealed that TUT4 and TUT7 uridylate mRNAs with short poly(A) tails, leading to 5′-to-3′ or 3′-to-5′ mRNA degradation by recruiting deadenylases, decapping enzymes, and exonucleases 20,30,31 . These studies indicate that oligo-uridylation is associated with RNA degradation, but how TUT4/7 uridylate RNAs remains unknown.
TUTs are recently shown to modulate inflammatory responses in mammalian cells. TUT4 uridylates and degrades IL-6-targeting miR-26a/b, leading to control of the quantity of Il6 mRNA in A549 cells following TNF treatment 24 . TUT7 is also known to modulate miR-26b uridylation and stability in IL-1β-stimulated chondrocytes through which positively regulates IL-6 expression 32 . A recent report revealed that TUT7 impedes the expression of a specific set of pro-inflammatory cytokines (including IL-6) in mice after challenge with Streptococcus pneumonia 33 . It is thus evident from in vitro and in vivo studies that TUT7 modulates inflammatory response. However, it remains an interesting question whether TUT7 can modulate inflammation by direct targeting mRNAs.
In this study, we reveal that TLR4 activation in macrophages upregulates TUT7 expression and TUT7 selectively regulates the expression of a subset of inflammatory cytokines. Both in vitro and in vivo studies show that TUT7 is critical for IL-6 and IL-12p40 upon LPS challenge and it is dependent on its nucleotidyltransferase activity. We further demonstrate that TUT7 directly controls Zc3h12a mRNA decay by uridylating its 3′ end, which subsequently prevents Regnase-1 from degrading Il6. Interestingly, Zc3h12a isolated from LPS-treated TUT7-sufficient bone marrow-derived macrophages (BMDMs) contains increased oligo-uridines (≥2 U) with short poly(A) tails at its 3′ end, whereas reduced oligo-uridylation is found in Zc3h12a transcripts from TUT7-deficient cells after the same treatment. Together, our results demonstrate that TUT7 functions as a regulator in TLR4driven inflammatory responses by mediating uridylation of and thus destabilizing the mRNAs of inflammatory mediators including Zc3h12a.
To assess the role of TUT7 in TLR4-driven inflammatory response, we knocked down its expression in murine RAW 264.7 macrophages by lentivirus-mediated shRNA transduction. Silencing Tut7 in RAW 264.7 macrophages significantly decreased Il6, Il12b, and Il1b mRNA, and cytosolic pro-IL-1β, and IL-6 and IL-12p40, but not Tnf and Nfkbia mRNA or TNF protein expression upon LPS challenge ( Supplementary Fig. 1d-h). Nevertheless, LPS-induced IκBα degradation, activation of MAPKs JNK and p38, and nuclear translocation of IRF3 were comparable in control and TUT7-silenced cells ( Supplementary Fig. 1i, j). These     Fig. 1b showed that TUT7 deficiency decreased the levels of Il6, Il12b, and Il1b mRNA and proteins (Fig. 1b, c, e) but not Tnf and Nfkbia mRNA or TNF protein in BMDMs response to LPS (Fig. 1d). Neither did TUT7 deficiency affect LPS-induced activation of IKK and MAPKs (Fig. 1f).
Tut7 +/+ and Tut7 −/− mice were challenged with a low-dose LPS. Results in Fig. 1h showed that the serum concentrations of IL-6 and IL-12p40 in Tut7 −/− were lower than in Tut7 +/+ mice after LPS challenge whereas the level of TNF remained comparable (Fig. 1h). These data clearly demonstrate that TUT7 is involved in TLR4-triggered inflammatory responses.
TUT7 regulation of Il6 mRNA stability requires its enzymatic activity. Aspartic acid residues at positions 1058 and 1060 on the TUT7 nucleotidyltransferase domain ( Supplementary Fig. 3) are critical for its nucleotidyltransferase activity 19 . To determine whether the nucleotidyltransferase activity of TUT7 is required for modulation of TLR4-driven cytokine production, we reconstituted TUT7-silenced RAW 264.7 macrophages with either human wild-type TUT7 (hTUT7) or a catalytically inactive mutant hTUT7(DADA) whose aspartate residues at positions 1058 and 1060 were replaced by an alanine ( Supplementary  Fig. 3). While reconstitution with wild-type TUT7 increased LPStriggered Il6 and Il12b, reconstitution with the hTUT7(DADA) mutant did not and Tnf expression was not affected by reconstitution with either wild-type hTUT7 or hTUT7(DADA) (Fig. 2a). Our results showed that nucleotidyltransferase activity is critical for the ability of TUT7 to modulate TLR4-driven cytokine response.
TUT7 and TUT4 are reported to control global mRNA turnover in HeLa cells 20 . Therefore, we hypothesized that TUT7 regulates inflammatory response in macrophages by modulating the stability of cytokine mRNAs. To address this possibility, we investigated Il6 as an example. The half-life of Il6 but not that of Tnf in TUT7-silenced macrophages was significantly reduced compared to that in Tut7 +/+ cells after stimulation with LPS (Fig. 2b). The 3′-UTR plays a critical role in controlling mRNA stability in innate immunity 8,10,34 . To determine whether TUT7 modulates Il6 turnover via its 3′-UTR, we transfected TUT7sufficient and -silenced RAW 264.7 cells with luciferase reporter constructs (pGL3) containing either mouse Il6 (mIl6) or human IL6 (hIL6) 3′-UTRs (Fig. 2c). The luciferase activity in TUT7depleted RAW 264.7 cells in response to LPS was significantly lower when the full-length mIl6-or hIL6-3′-UTR but not Tnf-3′-UTR was transduced compared to that in control cells (Fig. 2d). In the meantime, both control and TUT7-depleted RAW 264.7 cells transduced luciferase reporter driven by hIL6 promoter variants, AGC and GGG 35 , had similar response to LPS ( Supplementary Fig. 4). These data together indicate that TUT7 regulates Il6 expression induced by LPS not at the transcription level but posttranscriptionally via its 3′-UTR.
Analysis using TargetScan Mouse revealed three putative miRNA (miR-223, miR-224 and miR-376a) sites on mIl6 3′-UTR 56-104 ( Supplementary Fig. 5b). We mutated each miRNA target site to study whether TUT7-mediated suppression was dependent on any of the three miRNAs. Our results showed that the suppressive effect of TUT7 was lost when the target site for miR-376a but not miR-223 or miR-224 was mutated (Supplementary Fig. 5c). However, miR-376a target site is only observed in mIl6 3′-UTR, but not in hIL6 3′-UTR, suggesting that TUT7 control of Il6 expression is not via miR-376a. Interestingly, the putative miR-376a binding site overlaps with the conserved stemloop structure. To determine whether the stem-loop structure on BMDMs from wild-type (Tut7 +/+ ) and Tut7 −/− mice were incubated with 100 ng/ml LPS for indicated times. TUT7 protein expression was analyzed by immunoblotting (a). Il6, Il12b, and Il1b mRNA expression was determined by RT-qPCR (b, d). The levels of indicated inflammatory cytokines in cultural media were analyzed by ELISA (c, d). The levels of pro-IL-1β, IκBα, and phosphorylation of MAP kinases were analyzed by immunoblotting (e, f). g Volcano plot of the changes of innate immune-related gene expressions in Tut7 −/− BMDMs after 4 h treatment with 100 ng/ml LPS against wild-type BMDMs. The transcriptome of BMDMs from wild-type and Tut7 −/− was compared to baseline. The x-axis indicates the logarithm of p values to the base 2 of the fold-change (FC) and the y-axis reveals the negative logarithm of that to the base 10. Black horizontal dash line indicates significance at 5% criteria (p value = 0.05). log 2 FC > +0.2 and log 2 FC < −0.2 indicate increase of transcript levels by >20% and decreased by >20%, respectively. Green and red dots denote transcripts related to innate immune responses that were significantly lower (n = 111) and higher (n = 68) in response to LPS, respectively. The innate immune-related genes that are significantly differentially expressed are listed in Supplementary Tables 2, 3. h 8-week-old mice were injected intraperitoneally with LPS (50 μg/kg). Whole blood was collected 2 h after LPS challenge, and the concentrations of IL-6, IL-12p40, and TNF in sera were determined by ELISA. Data (except g) presented are representative of three independent experiments with triplicates in each experiment (error bars, mean ± S.D.). The p values were obtained from two-tailed Student's t test and are shown in the figure if p < 0.05. Source data are provided as a Source Data file.
Il6 is responsible for TUT7-mediated effects, we generated luciferase reporter constructs with either altered or similar stem-loop structures as the wild-type (Fig. 2g). Deletion of the internal loop structure (Δloop) abolished the suppressive effect of TUT7 (Fig. 2g, h). In addition, m1 and m2 mutants (with altered stem structures) both dampened TUT7-modulated inhibition of the luciferase activity; whereas m(1 + 2) mutant with an intact stem structure that was restored by combining both mutations regained the responsiveness of TUT7 (Fig. 2i). Taken together, TUT7-regulated Il6 mRNA destabilization is dependent on the stem-loop structure in its 3′-UTR 56-104 .
TUT7 controls Il6 mRNA stability through Regnase-1. Previous studies showed that TUT7 uridylates target RNAs to modulate their stability 20,25,26 . However, our RNA-immunoprecipitation (RNA-IP) data demonstrated that TUT7 did not associate with Il6 transcripts (Fig. 3a). Regnase-1 is reported to bind and control Il6 stability via the conserved stem-loop structure in Il6 3′-UTR 56-104 37 . Therefore, we hypothesized that TUT7 controls Il6 mRNA destabilization through regulating Regnase-1 expression. Results showed that the levels of mRNA and protein of Regnase-1 increased in LPS-stimulated TUT7-depleted RAW 264.7 cells (Fig. 3b, c) and BMDMs (Fig. 3d, e) compared with that in control cells. Regnase-1 belongs to the Zc3h12 family that is characterized by its conserved CCCH-type zinc-finger 15 . The mRNA level of another family member Zc3h12c was not altered in LPS-treated cells, regardless of the presence or absence of TUT7 ( Supplementary Fig. 6), suggesting a specific regulatory role of TUT7 in Zc3h12a expression. The half-life of Zc3h12a was longer (30 vs. 80 min) in TUT7-deficient RAW 264.7 macrophages than control cells upon LPS challenge (Fig. 3f). In addition, LPS-induced Zc3h12a expression was reduced in TUT7depleted cells reconstituted with wild-type TUT7, but not with hTUT7(DADA) mutant ( Fig. 3g), indicating that TUT7-mediated Zc3h12a expression after LPS stimulation depends on its nucleotidyltransferase activity. To further confirm TUT7mediated TLR4-induced cytokine production is through Regnase-1, we depleted Regnase-1 in Tut7 −/− BMDMs by lentivirus-mediated shRNA and challenged these cells with LPS. Deletion of Regnase-1 in Tut7 −/− BMDMs significantly increased Il6 and Il12b expression in response to LPS (Fig. 3h). However, LPS-induced Tnf and Nfkbia were similar in Tut7 −/− BMDMs expressing control and Zc3h12a shRNA (Fig. 3h). These data together demonstrate that TUT7 regulates TLR4-induced Il6 and Il12b expression by suppressing Regnase-1 expression.
We then determined whether TUT7 controls Zc3h12a mRNA stability via its 3′-UTR using luciferase reporter assay. We generated a set of luciferase reporter constructs containing full-length or different truncation of the Zc3h12a 3′-UTR and analyzed their luciferase activities in control and TUT7-depleted RAW 264.7 cells (Fig. 4a). The luciferase activity of full-length Zc3h12a 3′-UTR was upregulated in LPS-stimulated TUT7-depleted RAW 264.7 cells compared to that in control cells ( Supplementary   Fig. 7). To further assess whether TUT7 directly regulates Zc3h12a expression via its 3′-UTR, luciferase reporter constructs together with either wild-type or enzymatic inactive TUT7 were co-expressed in human embryonic kidney 293 (HEK 293) cells.
Like most mRNA 3′-UTRs, full-length mouse Zc3h12a 3′-UTR decreased luciferase activity compared to luciferase reporter control (Fig. 4b). This repressive effect was further enhanced by co-expression of wild-type TUT7, but not the hTUT7(DADA) mutant, suggesting that TUT7 directly regulates Zc3h12a in a nucleotidyltransferase activity-dependent manner (Fig. 4b).
TUT7 binds and uridylates Zc3h12a. Our results in Fig. 3g revealed that TUT7 directly modulated Zc3h12a mRNA stability through its 3′-UTR in an uridyltransferase activity-dependent manner. TUT7 possesses the ability to uridylate both mRNA and miRNA and to facilitate their degradation 20,23 . We thus hypothesized that TUT7 uridylates and degrades Zc3h12a after TLR4 engagement. To test this possibility, we first did RNA-IP to assess the association of TUT7 and Zc3h12a. Results in Fig. 5a, b showed that TUT7 was associated with Zc3h12a by 2 h after LPS treatment.
To further investigate whether TUT7 directly binds to Zc3h12a stem-loop structure, we performed RNA electrophoresis mobility shift assay (EMSA) using synthesized RNA probes. In line with the results of RNA-IP, TUT7 bound to murine Zc3h12a stemloop, but not that of Il6 (Fig. 5c). A previous study showed that the length of poly(A) tails negatively correlates with the frequency of uridylation 20 . Interestingly, TUT7 bound to Zc3h12a stemloop regardless of the length of poly(A) tails ( Supplementary   Fig. 2 Regulation of Il6 mRNA stability by TUT7 is dependent on its 3′-UTR 56-104 . a Tut7-silenced RAW 264.7 macrophages were reconstituted with either flag-tagged human wild-type TUT7 (hTUT7) or activity-dead TUT7 (hTUT7(DADA)) and challenged with 100 ng/ml LPS for 4 h. The expression of IL-6, IL-12p40, and TNF mRNAs and TUT7 protein were analyzed by RT-qPCR and immunoblotting, respectively. b RAW 264.7 macrophages expressing shControl and shTut7 were treated with 100 ng/ml LPS for 4 h followed by incubation with 5 μg/ml actinomycin D for the indicated times. Total RNAs were extracted and analyzed by RT-qPCR. c Schematic diagram of the firefly luciferase reporter constructs containing full-length (FL) and truncated forms of mouse (red line) or human (blue line) Il6 3′-UTR. d-f Control and Tut7-depleted RAW 264.7 cells were co-transfected with the indicated firefly luciferase reporter plasmids containing various mouse or human Il6 3′-UTR (mIl6 3′-UTR or hIL6 3′-UTR) or Tnf 3′-UTR and TK-renilla control reporter plasmid. At 48 h post-transfection, cells were treated with 100 ng/ml LPS for 8 h and the luciferase activities were determined. g Schematic diagram of the predicted stem-loop structure in human and mouse Il6 3′-UTR, and mutants whose stem-loop structure (m1 and m2) or loop structure (Δloop) were disrupted or deleted. The stem-loop structure of mIl6 3′-UTR or hIL6 3′-UTR was predicted by ref. 14 . Mutant m(1 + 2) was generated to swap sequences on the 5′ and 3′ sides of the stems without disrupting the stem-loop structure in mIl6 3′-UTR. h, i Control and shTut7-expressing RAW 264.7 cells were co-transfected with the indicated luciferase reporter plasmids and TK-renilla control reporter plasmid for 48 h. Cells were stimulated with 100 ng/ml LPS for 8 h and harvested for the analysis of luciferase activity. Data presented are representative of three independent experiments with triplicates in each experiment (error bars, mean ± S.D.). The p values were obtained from two-tailed Student's t test and are shown in the figure if p < 0.05. Source data are provided as a Source Data file. Fig. 8c), suggesting that TUT7 associating with Zc3h12a is dependent on its stem-loop structure rather than the length of its poly(A) tails. Furthermore, the association of TUT7 and Zc3h12a was dependent on its stem-loop structure rather than its sequence as the addition of RNA oligomers containing wild-type Zc3h12a or St(1 + 2)m, but not St1m nor St2m, stem-loop mutant abolished their binding (Fig. 5d). To assess whether TUT7 is able to uridylate Zc3h12a, we set up in vitro uridylation assay   using immunopurified TUT7 from HEK293T cells overexpressing Flag-tagged TUT7. RNA substrates were in vitro transcribed Zc3h12a or Tnf containing 3′-UTR (Fig. 5e). Interestingly, regardless of the coding sequences (CDSs) of the substrates, TUT7 was only capable of uridylating transcripts containing Zc3h12a 3′-UTR, but not those without 3′-UTR or those with Tnf 3′-UTR (Fig. 5f). Consistently, TUT7 did not uridylate transcripts containing Il6 3′-UTR albeit with the CDS of Zc3h12a (Fig. 5g). Furthermore, TUT7 preferentially uridylated transcripts containing Zc3h12a stem-loop rather than that containing Il6 stem-loop (Fig. 5h). Pull-down and immunoblotting experiments showed that Regnase-1 was not co-immunoprecipitated with TUT7 ( Supplementary Fig. 8a, b), which excludes the possibility of the involvement of Regnase-1 in TUT-mediated uridylation. These data together support the notion that upon LPS stimulation TUT7 directly binds and uridylates Zc3h12a 3′-UTR. We further examined Zc3h12a 3′-terminal uridylation in Tut7 +/+ and Tut7 −/− BMDMs using gene specific TAIL-Seq 31 (Fig. 6a). As expected, LPS significantly increased the frequency of uridylation on Zc3h12a, especially that of oligo-uridylation (≥2 U) (Fig. 6b). Deleting TUT7 significantly decreased uridylation on Zc3h12a (12% mono-uridine and 14% oligo-uridines among 1714 reads) after LPS stimulation compared to that of control (14% mono-uridine and 24% oligo-uridines among 1067 reads) (Fig. 6b). However, there is no observable difference in uridylation on Il6 mRNA between Tut7-deficient macrophages (9% mono-uridine and 5% oligo-uridine among 1730 reads) and control (9% mono-uridine and 6% oligo-uridine among 1756 reads) in response to LPS (Supplementary Fig. 9). Consistent with published studies 20,31 , the frequency of oligo-uridine (≥2 U) addition to Zc3h12a 3′ end negatively correlated with the length of poly(A) tails in wild-type cells but not in TUT7-deficient cells upon LPS challenge (Fig. 6c). Furthermore, TUT7 deficiency did not affect non-templated addition of cytosine and guanine to the 3′ end of Zc3h12a (Fig. 6d). Together, our results demonstrate that TUT7 binds Zc3h12a in macrophage stimulated by LPS and regulates Zc3h12a expression by uridylating its 3′ tail.
TUT4 and TUT7 regulate Zc3h12a expression through different mechanisms. TUT7 and TUT4 are highly homologous among members of the TUTase family. They have been reported to share various similar biological functions 38 . We thus examined whether TUT4 like TUT7 possesses the modulatory function of TLR4-driven effect through regulating Zc3h12a expression. We first assessed TUT4 expression in Tut7-deficient cells and found that the levels of TUT4 mRNA and protein were similar in TUT7-silenced and control cells (Supplementary Fig. 10a, b). However, unlike TUT7, depletion of TUT4 only marginally reduced Regnase-1 mRNA and protein expression after LPS stimulation ( Supplementary Fig. 10c, d), indicating that TUT4 and TUT7 participates in TLR4-triggered inflammatory responses through different mechanisms.

Discussion
Inflammatory cytokines are quickly induced upon infection in order to eliminate pathogens and repair damaged tissues. To prevent their detrimental effect on the host, their production must be tightly regulated. PTR is a quick and effective mechanism to modulate the expression of inflammatory cytokines 6 . The involvement of PTR in regulation of cytokine production during innate immune response is recently emerged. As shown in Fig. 7, we reveal that terminal uridyltransferase TUT7 expression is induced by TLR4 ligand LPS, and, it in turn participates in the regulation of the expression of a subset of cytokines, including IL-6 and IL-12β. TUT7 positively regulates Il6 mRNA stability through controlling Zc3h12a expression by its terminal nucleotidyltransferase activity after TLR4 activation. We further demonstrate that upon TLR4 engagement TUT7 directly binds and uridylates Zc3h12a to downregulate its expression, resulting in stabilization of Il6 mRNAs. Our results together demonstrate that TUT7 is a modulator of TLR4-triggered inflammatory response (Fig. 7). TUT7 fine-tunes the expression of IL-6, and we speculate other inflammatory cytokines as well, through directly uridylating Zc3h12a transcript.
Accumulating data have revealed that a number of TLRsinducible proteins play a role in modulation of inflammatory responses 39 . In this study, we extend this list to include TUT7. Our data indicate that TLR4-induced TUT7 expression requires IKK activity, and p38, but not JNK nor ERK, MAPK also contributes to its expression. These results suggest that transcription factors downstream of IKK and p38 MAPK may be required for TLR4-induced TUT7 expression. We and others have previously reported that p38 MAPK modulates the expression of certain LPS-induced genes via several transcription factors, including CCAAT/Enhancer binding protein β (C/EBPβ) and cAMP response element-binding protein (CREB) 40,41 . Interestingly, online prediction tool PROMO (http://alggen.lsi.upc.es/cgi-bin/ promo_v3/promo/promoinit.cgi?dirDB=TF_8.3) identified the conserved binding motifs for C/EBPβ and NF-κB on both human and mouse TUT7 promoters although it remains to be determined whether p38 MAPK regulates TLR4-triggered TUT7 expression through C/EBPβ.
Regnase-1 is an RNase. By controlling the degradation of target mRNAs, such as Il6 and Il12b, Regnase-1 is crucial for restraining inflammatory response during TLRs and IL-1 receptor (IL-1R) activation 37 . Therefore, the dynamics of Regnase-1 expression following TLRs and IL-1β stimulation ensures timely regulation of cytokine production. Accumulating data revealed that both PTR and PTM contribute to the regulation of Regnase-1 expression upon TLRs-and IL-1R activation 37 . Regnase-1 recognizes the conserved stem-loop structure in its 3′-UTR to Fig. 3 TUT7 decreases Zc3h12a mRNA stability in response to LPS challenge. a RAW 264.7 cells were treated with 100 ng/ml LPS for 2 h and cell lysates were incubated with control IgG or protein A-agarose beads-conjugated anti-TUT7 antibody at 4°C for 4 h. TUT7-interacting RNAs were extracted followed by RT-PCR (upper panel). Immunoprecipitation of TUT7 was examined by immunoblotting (lower panel). b-e Control (shControl) and Tut7knockdown RAW 264.7 cells and wild-type and Tut7 −/− BMDMs were treated with 100 ng/ml LPS for different periods of time. The expression of Regnase-1 protein (b, d) and Zc3h12a mRNA (c, e) were analyzed by immunoblotting and RT-qPCR, respectively. f Control-and shTut7-expressing RAW 264.7 cells were treated with 100 ng/ml LPS for 2 h followed by treatment with 5 μg/ml of actinomycin D for the indicated periods of time. Regnase-1 mRNA expression was analyzed by RT-qPCR. g Tut7-knockdown RAW 264.7 cells were reconstituted with either hTUT7 or hTUT7(DADA) and challenged with 100 ng/ml of LPS for 2 h. Cells were collected, and the expression of the indicated proteins and mRNAs were analyzed by immunoblotting and RT-qPCR, respectively. h Control-and shZc3h12a-expressing Tut7 −/− BMDMs were stimulated with 100 ng/ml LPS for the indicated periods of time. Total RNAs were prepared and the mRNA expression of the indicated genes was measured by RT-qPCR. Data presented are representative of three independent experiments with triplicates in each experiment (error bars, mean ± S.D.). The p values were obtained from two-tailed Student's t test and are shown in the figure if p < 0.05. Source data are provided as a Source Data file. degrade its own mRNA at the early phase of an inflammatory response 11 . Interestingly, Regnase-1 protein is quickly phosphorylated and degraded through IKK complex upon TLRs and IL-1R activation 11 . It may thereby unable to effectively eliminate its own mRNA, impeding the subsequent cytokine production.
Our findings demonstrate that TUT7 induces Zc3h12a degradation through uridylation on its 3′ end in the early phase of TLR4triggered inflammatory response. Our data together with a previous study 11 indicate that at least two distinct PTR mechanisms triggered by TLR4 destabilize Zc3h12a to promote the production   through uridylation, especially oligo-uridylation (≥2 U) as oligouridylation is crucial for mRNA decay 20,31 . Both TUTs also function redundantly to eliminate both maternal mRNA during early embryogenesis and viral RNAs in mammalian cells 28,43,44 .

CDS D A D A W T D A D A W T D A D A W T W T D A D A W T D A D A W T D A D A
Our results show that upon TLR4 activation, TUT7, but not TUT4, is upregulated and TUT7 subsequently associates with and oligouridylates Zc3h12a to regulate cytokine production. On contrast, TUT4 positively regulates Zc3h12a expression response to LPS through a yet-to-be investigated mechanism. TUT7 and TUT4 were reported to regulate Il6 expression through uridylation of miR-26, yet miR-26 uridylation is pervasive 24,32 . Our study demonstrates that mRNA uridylation can be specific and   inducible and that TUT7 is functionally distinct from TUT4 in innate immune responses. However, it is worth noting that LPSinduced Zc3h12a oligo-U-tails is not completely diminished in Tut7-deficient cells, indicating the involvement of additional TUT in Zc3h12a uridylation response to LPS. It is still unclear how TUT4 and TUT7 discriminate and mediate uridylation on the 3′ end of their target mRNAs. Generally, it is proposed that when the poly(A) tails of the target mRNAs become shorter than 20 nts, poly(A)-binding proteins (PABPs) are dissociated from the poly(A) tails, thereby allowing the binding of TUTs to mRNAs for subsequent uridylation 31,45 . This model is based on the assumption that uridylation by TUTs is an activity downstream of deadenylation but it excludes the possibility that TUTs may participate in deadenylation. In addition, it remains to be determined whether any cis-element on 3′-UTR is required for TUTs-mediated uridylation. Our data reveal that TUT7 directly binds to Zc3h12a through the conserved stemloop structure in its 3′-UTR and catalyzes uridylation, which probably marks the mRNA for degradation.
Regnase-1 is also known to target the same conserved stemloop structure as TUT7 does to facilitate its own mRNA decay 11,16 . It is unclear why Regnase-1 can recognize the stemloop structures appearing in both Zc3h12a and Il6 3′-UTRs, whereas TUT7 only interacts with Zc3h12a, but not Il6, stem-loop structure. Structural and functional studies suggest that Regnase-1 prefers binding to the stem-loop structure with a UAU loop and 5 to 9 base pairs in the stems 46 . Interestingly, the stem-loop structures in Il6 and Zc3h12a 3′-UTR are distinct: Il6 stem-loop structure harbors a typical Y-R-Y loop and 9 base pairs in the stem, whereas the one in Zc3h12a 3′-UTR is composed of a 5nucleotide loop and 5 base pairs in the stem 16 . A recent study employing Hamiltonian scores to predict how TUT7 distinguishes its target pre-let-7a from other unrelated RNAs suggests a critical role of the loop, but not the stem, structure for TUT7 binding 47 . These data together suggest that TUT7 may prefer to interact with the loop in Zc3h12a rather than Il6 3′-UTR, but this notion needs confirmation. In addition, Lin28 is shown to recognize a GGAG loop and subsequently recruit TUT4 to pre-let-7 48 . UPF1 helicase is also reported to cooperate with Regnase-1 to control Zc3h12a destabilization 16,49 . Therefore, we cannot exclude the possibility that an unknown RNA-binding molecule associates with the stem-loop structure of Zc3h12a and enlists TUT7.
Oligo-uridylation on mRNAs in mammalian cells is considered to be a tag for RNA degradation 20,29 . Addition of uridine to the 3′ end of RNAs by TUT7 and TUT4 induces 5′-to-3′ and 3′-to-5′ degradation by the XRN family of exoribonucleases and the exosome complex, respectively 20,23 . The 3′-to-5′ exoribonuclease Dis3L2 has been shown to form complex with TUT4/7 to degrade cytosolic noncoding RNAs, independent of the exosome 50 . However, recent studies indicate that Dis3L2 only plays a modest role in human cells to degrade global mRNAs with uridylated tails 51 . Rather it is a predominant exoribonuclease to eliminate uridylated mRNA during apoptosis 51 .
Here we reveal that LPS-induced TUT7 uridylates Zc3h12a and accelerates its mRNA degradation thereby promotes the production of its target cytokines. It appears that this regulatory mechanism to suppress Regnase-1 expression is at work in the early phase of inflammation when mRNAs of cytokines, such as IL-6 and IL-12β, are being actively translated. Inflammation is a highly dynamic and intricate process. Multiple factors collaborate to spatially and temporally control the production of inflammatory mediators at the levels of transcription, PTR, or PTM. A recent study demonstrated that Regnase-1 functions to degrade translationally active cytokine mRNAs during inflammation 16 . In addition, our transcriptome analysis revealed that 111 and 68 innate-related genes were up-and down-regulated, respectively, in Tut7 −/− cells upon TLR4 activation. TUT7 therefore may act in concert with other molecules to fine tune the expression of different sets of cytokine mRNAs. How different molecules and mechanisms work together to elaborately control inflammatory response is worth further investigation.
In summary, our study uncovers TUT7-mediated Zc3h12a uridylation as a posttranscriptional mechanism in regulation of TLR4-driven inflammatory cytokine response. This molecular mechanism involving TUT7-restricted Regnase-1 expression may Fig. 7 Model for the regulation of TUT7-mediated Regnase-1 expression. In response to TLR4 activation, IKK complex phosphorylates IκBα to promote NF-κB-driven transcription. Messenger RNAs of IL-6, IL-12β, Regnase-1, and TUT7 are upregulated. TUT7 targets and uridylates Zc3h12a on the stem-loop structure of Zc3h12a 3′-UTR. Uridylated Zc3h12a transcripts are eliminated by exonuclease, probably Dis3L2, leading to upregulation of the expression of a subset of cytokines, including IL-6 and IL-12β, at the early stage during TLR4-triggered inflammation. When TUT7 is depleted, elevated Regnase-1 results in a substantial decrease in IL-6 and IL-12β expression. The dashed and solid lines indicate indirect and direct pathways, respectively.
contribute to the generation of a specific set of inflammatory cytokines to shape a beneficial inflammation. Given the prominent role of IL-6 in inflammation, the TUT7/Regnase-1 axis pathway may also contribute to the pathogenesis of many inflammatory diseases.

Methods
Mice. Tut7 −/− mice were generated by Clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9-mediated genomic editing system in the Transgenic Mouse Models Core Facility of the National Core Facility Program for Biotechnology. Tut7 +/− mice were generated by targeting two sgRNAs to the first and 26th introns of the Tut7 gene on C57BL/6 genetic background. All animals were bred in the specific-pathogen-free animal facility. Mouse experiments were carried out in accordance with animal welfare guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) of the College of Medicine, National Taiwan University (approval no. 20150507).
Cell cultures and preparation of bone marrow-derived macrophages. RAW 264.7 macrophages were cultured in RPMI 1640 (Gibco, Pittsburgh, PA) supplemented with 10% (vol/vol) heat-inactivated fetal bovine serum (FBS) and 100 U/ml penicillin/streptomycin. HEK293T and HEK 293 cells were cultured in DMEM containing 10% FBS and 100 U/ml penicillin/streptomycin. All cells were maintained at 37°C in a humidified 5% CO 2 atmosphere. BMDMs were prepared as previous description 5 . Briefly, femurs and tibia bones were collected from 6 to 8week-old mice and bone marrow was flushed out with DMEM medium using a 25gauge syringe. The bone marrow progenitor cells were harvested, and differentiated in high-glucose DMEM medium containing 20% L929 cell-conditioned medium for 7 days. Adherent BMDMs were collected and cultured in DMEM containing 10 ng/ml M-CSF overnight for further experiments.
RNA extraction and Real-time quantitative RT-PCR. Total cellular RNA was isolated using NucleoZol (MACHEREY-NAGEL, Düren, Germany). One μg of total RNA was used to synthesize cDNA with the RevertAid H Minus First-Strand cDNA Synthesis Kit (Thermo Scientific, Rockford, IL) following the manufacturer's instructions. The amount of cDNA was determined by Real-time quantitative PCR (RT-qPCR) using Maxima ® SYBR Green/Fluorescein qPCR Master Mix (#4367659, Thermo Scientific, Rockford, IL) according to the manufacturer's instructions. All RT-qPCR values of interesting genes were normalized to cyclophilin A transcript as an internal control. All data were presented as fold-change relative to the unstimulated sample. The primer sequences are listed in Supplementary Table 5.
mRNA stability assay. RAW 264.7 macrophages were challenged with 100 ng/ml LPS for 4 h, and then incubated with actinomycin D (5 μg/ml) for the indicated times. The cells were collected and total cellular RNAs were prepared for mRNA quantification by RT-qPCR as described above.
Dual-luciferase reporter and enzyme-linked immunosorbent assays. Cells were co-transfected with the indicated firefly luciferase reporter plasmid, pRL-TK-renilla luciferase plasmid, and TUT7 or empty control plasmid, and cultured for 48 h. Cells were harvested and lysed, and firefly and renilla luciferase activities were determined using the Dual-Luciferase reporter assay system (Promega, Madison, WI) according to the manufacturer's instructions. The levels of cytokines in cultural supernatants were measured by DuoSet ELISA systems (R&D Systems, Minneapolis, MN) following the manufacturer's instructions.
In vitro transcription and In vitro uridylation. In vitro transcription was carried out by T7 RNA polymerase (Promega, Madison, WI) according to the manufacturer's instructions. Briefly, 5 μg of linearized DNA template was incubated with 40 units T7 RNA polymerase in 100 μl reaction mixtures containing 40 mM Tris, pH 7.9, 6 mM MgCl 2 , 2 mM spermidine, 10 mM NaCl, 10 mM DTT, 2.5 mM NTP mix, 100 units RNase inhibitor, for 2 h at 37°C. RNA transcripts were then purified by NucleoZol, precipitated with 100% isopropanol (Sigma-Aldrich, St. Louis, MO), and resolved in 10 μl RNase-free water. For in vitro uridylation assay, 1 μg in vitro synthesized RNAs were incubated with wild-type or enzyme-dead TUT7 immunoprecipitated from HEK293T cells transfected with TUT7 expression plasmids and α-32 p-UTP for 1 h at 37°C. RNA samples were separated on a 1% formaldehyde-agarose gel electrophoresis and visualized by SYBR Green II staining and radiography.
RNA-immunoprecipitation. Cells were harvested and lysed in APB lysis buffer containing protease inhibitors and RNase inhibitor. 5% of cell lysates were kept as the inputs while remaining lysates were incubated with the appropriate antibody and Protein A-conjugated agarose beads for 4 h at 4°C. The beads were pelleted and washed three times with APB containing RNase inhibitor. 5% of the beads were subjected to SDS-PAGE to validate the efficiency of the immunoprecipitation. The remaining beads and the inputs were resuspended in 200 μl NucleoZol for RNA extraction followed by RT-PCR and RT-qPCR as mentioned above. The primer sequences, Zc3h12a_3′-UTR Forward 5′-ATCACAGATAGCGGTCCCCA-3′, Zc3h12a_3′-UTR Reverse 5′-GGCAATAGCTTTTTTTTTCTTTTAA-3′, Il6 Forward 5′-ACAAGAAAGACAAAGCCAGAGTC-3′, and Il6 Reverse 5′-ATTG GAAATTGGGGTAGGAAG-3′, were used in RT-PCR and RT-qPCR.
RNA sequencing. Total RNAs harvested from BMDMs were purified by Trizol and subjected to next-generation sequencing and data analysis (Changgong Biotechnology, Taipei, Taiwan) following the manufacturer's instructions. Briefly, the total RNAs were isolated by poly-dT magnetic beads, fragmented, reverse transcribed using random primers, and synthesized the second strand using linkers containing barcoded to generate the cDNA libraries. The cDNA libraries were amplified by PCR for the indicated cycles and the specific size products were selected and purified before sequencing on an Illumina HiSeq PE150 for 75 nts of each read. Total of 4.45-11.8 million reads (Reads Per Kilo bases per Million reads, RPKMs) were obtained from each sample. RPKMs were compared to mouse RefSeq-RNA mm10 and quantitated using FANSe3 (Fast and Accurate mapping algorithm for Next-generation Sequencing, the 3rd generation 53,54 ). The differential genes expression between control and LPS stimulation of each genotype (wild-type and Tut7 −/− ) was quantified using edgeR 55 and FANSe3. To identify the differential genes expression in response to LPS, fold-change (>1.2X) and t tests (p value < 0.05) of Tut7 −/− BMDMs at 4 h after LPS treatment were compared to that of Tut7 +/+ BMDMs. All detailed analysis was consulted by Next-generation Sequencing Analysis Cloud System (Chi-Cloud). The differentially expressed innate immune-related genes are listed in Supplementary Table 2 (downregulated  genes) and Supplementary Table 3 (upregulated genes). The NCBI GEO accession number for this experiment in present paper is GSE136161.
TAIL-Seq of specific genes. Total RNAs collected from BMDMs were purified by NucleoZol and subjected to 3′-linker ligation with adenylated linker-1, rRNA depletion, and cDNA library construction. Briefly, linker-1 (CTGTAGGCACCA TCAAT) was firstly adenylated with Mth RNA ligase (#M2611A, New England BioLabs, Ipswich, MA), and adenylated linker-1 was ligated to the extracted RNAs. The 3′-linker-ligated RNAs were further depleted of rRNA by NEBNext ® rRNA Depletion Kit (E6310S, New England BioLabs, Ipswich, MA) and fragmented by RNase T1 (EN0541, Thermo Scientific, Rockford, IL). Ribosomal RNA-free transcripts were reversely transcribed using SuperScript™ III First-Strand Synthesis System (Thermo Scientific, Rockford, IL) with primer complement to linker-1 with partial illumina sequence (5′-ACACTCTTTCCCTACACGACGCTCTTCCGATC TATTGATGGTGCCTACAG-3′). The 3′ end of individual transcript was amplified with gene specific primer and primer complement to linker-1 (TTTAAATGAAA AAGGTTGACAAAATAAA for Zc3h12a-forward primer and TGTTTAGACTGT CTTCAAACAAATAAA for Il6-forward primer, and RT linker-1 for reverse primer). Then the second-run PCR was used to add all the necessary sequences that were required for illumine sequencing using Q5 ® High-Fidelity DNA Polymerase (M0491S, New England BioLabs, Ipswich, MA). The amplified libraries were purified using 1X Sera-Mag Select magnetic beads (29343045, Cytiva/GE Healthcare, Little Chalfont, UK). The specific size (200-400 bps) products were selected, purified, and mixed with 20% (of total libraries) PhiX control library before sequencing on an Illumina HiSeq 2500 Rapid v2, Paired End 2*150 for 150 nts of each read. Total of 0.3-1.4 million reads were obtained from each sample. The reads were mapped to mouse RefSeq-RNA mm10 using HISAT2, and further analysis of Zc3h12a and Il6 was trimmed by 5′ and 3′ Illumina adapter sequences and filtered using Filter FASTA program to filter out Zc3h12a and Il6 by linker-1 sequence and the specific criteria (TTTAAATGAAAAAGGTTGACAAAATAAA for Zc3h12a and TGTTTAGACTGTCTTCAAACAAATAAA for Il6) on Galaxy bioinformatics online tool (https://usegalaxy.org/). The NCBI GEO accession number for this experiment in present paper is GSE164259. The oligonucleotides used in TAIL-Seq were listed in Supplementary Table 6.
RNA electrophoretic mobility shift assay and competition assay. Cy3-labeled and non-labeled RNAs listed in Supplementary Table 7 were purchased from Genomics (New Taipei City, Taiwan). Seventy five fmole synthesized RNAs were incubated with immunopurified TUT7 from HEK293T cells at 37°C for 30 min, and RNA samples were then mixed with RNA loading dye (1× TBE, 20% glycerol, 0.05% bromophenol blue) and separated on a 6% native PAGE, followed by detection of fluorescence signals from Cy3-fluorophore or SYBR Green II staining using the iBright FL1000 Imaging System (Invitrogen, Carlsbad, CA).
Statistical analysis. The results are presented as mean ± SD. The differences between two groups were determined by two-tailed Student's t test. The exact p values are shown in the figures if p is <0.05, which is considered statistically significant.

Data availability
RNA-Seq and TAIL-Seq data have been deposited into NCBI GEO under the accession number GSE136161 and GSE164259, respectively. The datasets of LPS-induced BMDMs and RAW 264.7 cells were obtained from GEO with the accession number of GDS5623. The data that support the findings of this study are available from the corresponding author upon reasonable request. Source data are provided with this paper.